research - Leicester Research Archive - University of Leicester

15 downloads 0 Views 131KB Size Report
applied to health,2 Stephen Sutton, professor of behavioural science,1 David P French, professor of health psychology,3 Joana Vasconcelos, reader in medical ...
RESEARCH Are people with negative diabetes screening tests falsely reassured? Parallel group cohort study embedded in the ADDITION (Cambridge) randomised controlled trial Charlotte A M Paddison, ESRC postdoctoral research fellow,1 Helen C Eborall, lecturer in social science applied to health,2 Stephen Sutton, professor of behavioural science,1 David P French, professor of health psychology,3 Joana Vasconcelos, reader in medical statistics,1 A Toby Prevost, senior medical statistician,4 Ann-Louise Kinmonth, professor of general practice,1 Simon J Griffin, assistant unit director5 1 General Practice and Primary Care Research Unit, University of Cambridge, Institute of Public Health, Cambridge CB2 0SR 2 Department of Health Sciences, University of Leicester, Leicester LE1 7RH 3 Applied Research Centre in Health and Lifestyle Interventions, Coventry University, Coventry CV1 5FB 4 Department of Public Health Sciences, King’s College London, London SE1 3QD 5 MRC Epidemiology Unit, Institute of Metabolic Science, Addenbrooke’s Hospital, Cambridge CB2 0QQ Correspondence to: C Paddison [email protected]

Cite this as: BMJ 2009;339:b4535 doi:10.1136/bmj.b4535

BMJ | ONLINE FIRST | bmj.com

ABSTRACT Objective To assess whether receiving a negative test result at primary care based stepwise diabetes screening results in false reassurance. Design Parallel group cohort study embedded in a randomised controlled trial. Setting 15 practices (10 screening, 5 control) in the ADDITION (Cambridge) trial. Participants 5334 adults (aged 40-69) in the top quarter for risk of having undiagnosed type 2 diabetes (964 controls and 4370 screening attenders). Main outcome measures Perceived personal and comparative risk of diabetes, intentions for behavioural change, and self rated health measured after an initial random blood glucose test and at 3-6 and 12-15 months later (equivalent time points for controls). Results A linear mixed effects model with control for clustering by practice found no significant differences between controls and people who screened negative for diabetes in perceived personal risk, behavioural intentions, or self rated health after the first appointment or at 3-6 months or 12-15 months later. After the initial test, people who screened negative reported significantly (but slightly) lower perceived comparative risk (mean difference −0.16, 95% confidence interval −0.30 to −0.02; P=0.04) than the control group at the equivalent time point; no differences were evident at 3-6 and 12-15 months. Conclusions A negative test result at diabetes screening does not seem to promote false reassurance, whether this is expressed as lower perceived risk, lower intentions for health related behavioural change, or higher self rated health. Implementing a widespread programme of primary care based stepwise screening for type 2 diabetes is unlikely to cause an adverse shift in the population distribution of plasma glucose and cardiovascular risk resulting from an increase in unhealthy behaviours arising from false reassurance among people who screen negative. Trial registration Current controlled trials ISRCTN99175498.

INTRODUCTION A national screening programme for cardiovascular risk factors, including testing for type 2 diabetes, is now being implemented.1 Justification for screening programmes requires that the overall benefit of testing—resulting mainly from early intervention in people at risk—is greater than any possible harms of testing the population. Whether this criterion is met for screening for diabetes remains uncertain.2 3 Research evaluating the benefits of screening and early intervention among people at risk of diabetes has shown that stepwise screening identifies those with a raised and potentially modifiable risk of coronary heart disease.4 5 Studies have also examined the direct harms to people attending screening and shown limited evidence of adverse psychological effects associated with diabetes screening programmes in terms of increased anxiety and depression.6-9 In part, this may be because stepwise screening in the context of primary care offers an opportunity for psychological adjustment as participants progress through the screening programme.10 Research describing the potential harms of screening has focused on people who screen positive,11 12 and few studies have considered the potential indirect harms to people attending screening. In particular, research has yet to examine whether people with a negative test result at diabetes screening may be falsely reassured.13 14 This is important because the potential adverse effects of false reassurance include decreased intentions for behavioural change,3 13 justification of unhealthy behaviours through a “certificate of health” effect,15 16 and delays in seeking medical help.17 As the vast majority (90%) of people who take part in diabetes screening programmes test negative, even a small harm to those who test negative (for example, through a reduction in intentions to be physically active) may outweigh a large benefit to the few people who test positive. The population distribution of plasma glucose in the United Kingdom is approximately normal and shows a positive linear relation to page 1 of 7

RESEARCH

cardiovascular risk.18-20 Consequently, if a substantial proportion of people who take part in screening for diabetes are falsely reassured and show a reduction in intentions to improve their diet or increase their level of physical activity as a result of screening, then the net effect of screening could be to move the population distribution of plasma glucose and cardiovascular risk in the wrong direction. The potential problem of false reassurance in cardiovascular screening has been described as “an unintentional adverse effect of screening . . . evident when people erroneously conclude that their screening result means that they are at less risk than they actually are.”21 Some evidence shows false reassurance associated with screening for cardiovascular risk factors,16 22 but no studies to date have examined false reassurance when quantified as a reduction in intentions for behavioural change or a decrease in perceived risk in the context of diabetes screening.13 14 In particular, a need exists to assess false reassurance among people with risk factors for diabetes who receive a negative test result at screening.23 This is especially timely given the Department of Health’s decision to include screening for diabetes in the national vascular risk assessment programme.1 We aimed to investigate false reassurance in the short term (after the initial appointment) and the medium to long term (three to six months and 12-15 months later) among participants who test negative for diabetes in a primary care based stepwise screening programme. We hypothesised that those who test negative for diabetes would report lower perceived risk of developing diabetes, lower intentions for behavioural change, and higher self reported health than a (non-screening) control group. METHODS Participants were registered at 15 primary care practices included in the Cambridge arm of the AngloDanish-Dutch Study of Intensive Treatment in People with Screen Detected Diabetes in Primary Care (the ADDITION trial). A comprehensive description of the screening procedure and recruitment methods for this study has been published.4 7 The ADDITION (Cambridge) trial aims to evaluate the effectiveness and cost effectiveness of a stepwise screening strategy for type 2 diabetes and intensive multifactorial treatment for people with screen detected diabetes in primary care.4 The primary outcomes in this trial are modelled cardiovascular risk at one year and cardiovascular mortality and morbidity at five years after diagnosis of diabetes. Embedded within the ADDITION (Cambridge) trial was a substudy that aimed to quantify the short term and long term psychological impact of primary care based stepwise screening for type 2 diabetes.7 Specifically, the psychological impact study aimed to examine the direct harms or benefits to screening attenders in terms of anxiety and depression,7 10 and to investigate the indirect harms caused by false reassurance (the focus of this paper). The psychological page 2 of 7

impact study was a controlled trial comparing screening attenders (n=4370) from 10 screening practices with controls (n=964) from five practices whose patients were not invited to screening; the primary comparison to test for false reassurance is between those who test negative for diabetes at screening and (non-screened) controls. The figure shows the design of this study and the flow of participants through the screening programme. In the 10 screening practices, patients aged 40-69 years identified by the Cambridge diabetes risk score as being in the top quarter of risk of having undiagnosed diabetes (n=6416) were invited to attend a stepwise diabetes screening programme.24 The stepwise screening programme has been described previously.4 7 The invitation letter informed patients that “from the information in your medical records you have been identified as being at risk of having undiagnosed diabetes,” and invited them to attend a random capillary blood glucose test in general practice. People who tested positive at this initial test (n=1787) were told: “According to this test, you have about a 15-20% chance of having diabetes. In order to find out for certain whether you have diabetes or not, we need to do further tests.” Participants with a random blood glucose of 5.5 mmol/l or more were invited to return on a different day for a fasting capillary blood

Attended random glucose test (n=4370) Screened negative (n=2583) Responders: 81%, 67%, 67% Screened positive (n=1787) Did not attend subsequent tests (n=129) Attended fasting glucose test (n=1658) Screened negative (n=1297) Responders: 67%, 72%, 69% Screened positive (n=361) Did not attend subsequent tests (n=65) Attended oral glucose tolerance test (n=296) Screened negative (n=146) Responders: 67%, 72%, 78% Screened positive: newly diagnosed diabetes (n=150) Responders: 66%, 74%, 75%

Control participants (n=964) Control group A (n=484) Responders: 54%, 43%, 37% Control group B (n=480) Responders: N/A, 50%, 43%

Flow of participants through screening programme, with questionnaire response rates at each time point (after initial appointment, 3-6 months later, 12-15 months later) BMJ | ONLINE FIRST | bmj.com

RESEARCH

glucose test. People who tested positive at the fasting blood glucose test were told: “This test result does suggest that you have diabetes, but a diagnosis of diabetes is for life, so it is very important that we make absolutely sure that you have the condition. To confirm diagnosis of diabetes you will need to make an appointment for a blood test at a local hospital.” Those with either a fasting blood glucose of 6.1 mmol/l or above, a fasting blood glucose of 5.5-6.1 mmol/l and a glycated haemoglobin of 6.1% or more, or a random blood glucose of 11.1 mmol/l or above were invited to attend for a standard 75 g oral glucose tolerance test at an outpatient centre. Diagnosis of type 2 diabetes was made according to the World Health Organization’s criteria.25 Participants who tested negative for diabetes at any stage in the stepwise screening programme were told: “This test is normal; you do not have diabetes.” In the screening practices, patients who attended for the initial test (n=4370) were given a questionnaire to complete and return by mail to the research centre. Questionnaire data from this time point (time 1) represent participants’ views after the initial test when they had been informed either that they did not have diabetes (screened negative) or that they needed further tests (screened positive). Follow-up questionnaires were sent three to six months (time 2) and 12-15 months (time 3) after the initial test. In the five control practices, a 25% random sample of participants identified as being in the top quarter of risk (n=964) were randomly allocated to two groups (A and B). Group A (n=484) were sent questionnaires at three time points equivalent to those for participants in the screening practices, and group B (n=480) received questionnaires at times 2 and 3 only—for a substudy investigating measurement effects.26 The letter accompanying the questionnaire informed control participants that “some people of your age can have diabetes without realising it.” Consent to participate in the questionnaire study was obtained during attendance at the initial test; control participants received an information sheet and consent form from their practice with their first questionnaire. At each time point, participants who had not returned a questionnaire after approximately three weeks were sent one reminder (including another copy of the questionnaire). Outcome measures Perceived personal risk—Participants were asked to estimate their chance of getting diabetes at some time in their life. Responses were given as a percentage score on an 11 point scale ranging from 0% to 100%. Perceived comparative risk—Participants were asked to indicate their chance of getting diabetes, compared with other people of their age. Response options provided on a five point rating scale were (1) much lower, (2) a little lower, (3) about the same, (4) a little higher, and (5) much higher.27 Behavioural intention—Participants were asked to rate three statements of intention to be more physically active, to eat a lower fat diet, and to reduce the amount BMJ | ONLINE FIRST | bmj.com

of dietary sugar in the next 12 months. Response options provided on a five point rating scale were (1) strongly disagree, (2) disagree, (3) neither agree nor disagree, (4) agree, and (5) strongly agree. Self rated health—A single item asked about the participant’s general health, with response options of (1) excellent, (2) very good, (3) good, (4) fair, and (5) poor. Sample size We determined the sample size in a previous substudy to quantify the effects of screening on anxiety and depression.7 The observed 95% confidence interval for the primary outcome, personal risk, in this study incorporates the observed clustering of the outcome by practice and provides the basis of an indicative post hoc calculation of detectable effect sizes. We would have 80% power to detect a difference between screened and control groups in mean personal risk of 6.5 scale units at baseline (standardised difference of 0.30 SD) and 5.75 units at later time points (equivalent to effect size differences of 0.3 SD and 0.27 SD). Analyses We first explored differences in the demographic and clinical characteristics of control group participants and screening attenders (and also those who screened negative) by using analysis of variance and χ2 tests. We used χ2 tests to assess differences between questionnaire responders and non-responders. Primary analyses used a linear mixed effects model to compare participants who screened negative for diabetes with controls on outcome measures at the initial time point and at three to six months and 12-15 months after screening. Because allocation of participants to groups for the controlled trial was practice based,4 7 we entered practice as a random effect in all multivariate models to account for potential clustering. We assessed hypotheses by using two sided tests at the 5% level of significance. Firstly, we assessed differences in perceived risk, behavioural intentions, and self rated health at time 1 between people who screened negative and controls. Secondly, we assessed differences in perceived risk, behavioural intentions, and self rated health between four screening subgroups and controls at times 2 and 3. To further examine the effect of progress through a stepwise screening programme, we used linear regression analysis to examine the relation between number of screening and diagnostic tests and perceived risk among people who screened negative for diabetes. RESULTS Of 6416 people invited to screening, 68% (n=4370) attended the initial random capillary blood glucose test in general practice. Among the screening attenders, 82% (n=1465) of 1787 screen positives and 81% (n=2092) of 2583 screen negatives returned a completed questionnaire after their first appointment. Removal of 310 questionnaires from people who screened positive that were returned late (after the date of a follow-up fasting capillary blood glucose test page 3 of 7

RESEARCH

or oral glucose tolerance test) gives an amended response rate of 65% (n=1162) for the screen positive group. Control group response rates were 54% (n=261) at the first contact point, 47% (n=454) at three to six months, and 40% (n=386) at 12-15 months. The figure shows the flow of participants through the screening programme and the questionnaire response rates at each time point. We found no significant differences between controls and screening attenders in terms of age, sex, body mass index, diabetes risk score, or prescription of antihypertensive drugs at baseline (table 1). No significant differences in baseline data existed between controls and people who screened negative at the initial test (table 1). Overall, at time 1, self rated health was high and intentions to change behaviour were modest: 77% of participants rated their health as good, very good, or excellent; 61% intended to reduce their dietary fat; and 54% intended to increase their levels of physical activity.

found a significant linear trend in mean scores for perceived personal risk at time 2 (P